A new system developed in the Bioimaging and Machine Vision laboratory at the University of Maryland automatically detects bone fragments, with an almost zero percent rate of misses, before packaging. Designed by assistant staff scientists Xin Chen and Hansong Jing, the system is the first to combine X-ray and laser range imaging technologies for inspecting poultry, a promising advance for the American industry that produces more than 13.2 billion pounds of boneless poultry annually.

"We began this project after learning that the existence of bone fragments in boneless poultry was the number one customer complaint of poultry companies," Chen said. Metal, glass and plastic bits also occasionally turn up in poultry meat.

Chicken Fingers

In many poultry processing plants, inspection of boneless poultry is still performed manually. Accuracy depends on the sensitivity of inspectors' fingers. Since the poultry is kept extremely cold to protect food quality, inspectors' fingers become numb quickly, making it difficult to find bone fragments. Warming the poultry would improve the working environment, but increase risk of bacterial growth in the meat.

"Limitations in manual inspection allow a small, but significant, number of bone fragments to be left in the boneless poultry after inspection," said Yang Tao, director of the Bioimaging and Machine Vision laboratory.

X-ray imaging, the most common bone fragment detection technology currently in use, misses up to 30 percent of the bone fragments, because it cannot adjust for uneven thickness of poultry. The uneven thickness produces false patterns in the X-ray image, making it difficult to distinguish between bone fragments and normal poultry.

Zero Percent Errors

The Maryland inspection system combines laser range imagers with X-rays, to get a more accurate look at hidden fragments. First, a uniform X-ray beam penetrates the poultry. When the X-ray encounters a bone fragment, a dark spot appears in the X-ray image, because the bone absorbs more of the X-ray beam than the meat does.

However, thicker areas of meat can also show up as dark areas, producing false patterns in the X-ray image. Adding the laser range imaging subsystem helps distinguish between bone fragment and thicker meat areas.

The scientists also designed a computer program that determines if a dark, suspicious region is a bone fragment or just meat. Because bone fragments vary significantly in terms of brightness and shape in the X-ray image, they can be classified. The combined system identifies various physical contaminations automatically while the chicken breasts are still on the conveyer belt.

"We have zero percent error for large and small bones, but some tiny ones slip through," said Jing. "The system can currently detect a bone fragment as small as a pinhead."

From Lab to Meat Case

According to the scientists, this new technology should not increase the price of chicken breasts at the grocery store. Instead, it allows for lower labor costs, an improved work environment and increased accuracy over manual inspection or X-ray imaging alone.

The researchers say the system is also safe for consumers. "The radiation from these X-rays is low energy, approximately the same as a mammogram," said Tao. "The X-rays and lasers cause no structural damage to the chicken breasts."

The system is still in the research and demonstration stage, but the researchers hope to put it to the test in a poultry processing plant. "Our demonstrations have been quite successful," Jing said. "We have performed demos for people including representatives from Perdue Farms; Allen Family Foods, a poultry processing plant; and Safeline, a packaging company. We expect the system to hit the poultry processing plants in the next year or two."

Chen and Jing's research is funded by grants from the USDA, the United States-Israel Binational Agricultural Research and Development Fund, and the Poultry and Egg Association.

University of Maryland College of Agriculture and Natural Resources - August 2004